View
12
Download
0
Category
Preview:
Citation preview
Math Enabled Innovation
Dr. Paul E. KrajewskiDirector, Vehicle Systems Research Lab
GM Global Research & Development
J. T. Wang, Bahram Khalighi, Yilu Zhang
Connected Living
Fuel Economy CO2
– Aero
– Electric Vehicles
– Lightweight Materials
Personal Security
Future / Summary
AGENDA
Mobile-cellular subscriptions approaching 7B, the number of people on the earth
3B use the Internet
Facebook has over 1.4 billion active users globally; 85% are mobile users
90% of the world’s data has been created in the last two years –2.5 quintillion bytes of data created every day!
CONNECTED LIVING
CUSTOMER EXPECTATIONS
Bring their digital life into their vehicle Bring their vehicle into their digital life
Vertical Implementation ICS BinArm Rest
Phone overheating in vehicle cabins using inductive wireless charging led to
unacceptable phone charging performance
WIRELESS CHARGING
Air gap for convective cooling
Phone
POWER Module
• Passive cooling incorporates an air-gap to facilitate the dissipation of trapped heat
from the interface between the phone and charging surface
• CFD tools like Fluent were used to design bumps and spacing
Armrest
PASSIVE COOLING
Eva
po
rat
or
ICS Bin
ICS Bin
2 mm air gap
Phone temperature
Air flow to ICS bin
ACTIVE THERMAL MANAGEMENT
Air from the HVAC module routed to the phone to maintain it within operational temperatures.
• Significant improvement in phone charge times
during extreme hot and cold.
• Increased customer acceptance and satisfaction
by not compromising charging performance.
• Enabled GM to be the first OEM to provide an
integrated dual protocol (WPC and PMA-
Powermat) wireless charging system (inductive) in
vehicles.
Phone Temperature
RESULTS
IHS/CSM
Segment
NHTSA
Segment
Representative
Model
Tailpipe
CO2
(g/KM)
Fuel
Economy
(mpg)
B-segment Car Compact Car Chevrolet Sonic 78 61.1
D-segment Car Mid-Size Car Buick Regal 88 54.9
E-segment Car Full-Size Car Cadillac XTS 102 48.0
C-segment Truck Small SUV Chevrolet Equinox 102 47.5
D-segment Truck Mid-Size CUV Cadillac SRX 112 43.4
FSFF Truck Large PU Chevy Silverado 150 33.0
2025 CO2 AND FUEL ECONOMY TARGETS
ELECTRIFICATION
11
Chevy Volt
Battery pack
BATTERY THERMAL MANAGEMENT
• Temperature impacts battery performance and life.
• Li-Ion batteries in hybrid and EV vehicles undergo constant charging and
discharging.
• Active battery cooling is necessary to maintain the cell temperatures within
allowable temperature limits (25oC ~ 35oC)
• Peak durability and reliability requires DT<5oC within the cell and across the pack
BATTERY PACK DESIGNElectrochemistry
Heat Transfer
Fluid Flow
Material Properties
Design
Curr
ent C
olle
cto
r (C
u)
+-
Curr
ent C
olle
cto
r (A
l)
sep
Negative
Ele
ctr
ode
Separa
tor
Positiv
e
Ele
ctr
ode
L
Li+
e-
rLi
xC
6
rLi
yCoO
2
e-
Electrolyte
x
cs,e
ce
cs(r)c
s(r)
cs,e
Computer-aided software design tools for hybrid/electric and
electric vehicle (HEV/EV) batteries to reduce cost, improve
performance and increase life.
• GM : End user requirements,
verification/validation, project management
• ANSYS : Software dev. and commercialization
• ESim : Cell level sub models, life model
• NREL : Technical monitor
Partners
Funding provided by DOE Vehicle Technologies Program .
CAEBAT
• Cooling design implemented on 2016 Chevy Volt
• Enabled more uniform temperature
• Lower power requirements for cooling
RESULTS
Total pressure drop
= 28.4 kPa
2011 Chevy Volt
Total pressure drop
= 12.4 kPa
2016 Chevy Volt
AERODYNAMICS
16
Parametric sensitivity
•Approximate a vehicle shape with many
design parameters, such as angle, length,
curvatures…
•Calculate the sensitivity for each design
variables
N design parameters
N+1 CFD Flow solutions
DRAG REDUCTION - CONVENTIONAL METHOD
• Incremental.
• Takes too long.
• Doesn’t account for multiple
interactions
Precept (Cd=0.19)
DRAG REDUCTION - ADJOINT METHOD
• CFD provides diagnostic information for drag reduction.
• CFD is not prescriptive on shape sensitivities nor direction of the shape change.
• Adjoint optimization method provides aerodynamic design information for both shape
sensitivity and direction for shape improvements.
18
•All the surface mesh (x,y,z)
locations are the design
parameters
•Large number of design
parameters 0.5 to 1 million
N design parameters
1 CFD Flow solution
1 Adjoint solution
CFD Flow Solver
Adjoint Solver
Calculate
Shape Sensitivity
Shape
Change
ADJOINT METHODOLOGY
19
Reduce the stagnation region
Recessing the A-pillar to reduce the transverse pressure gradient
Lowering the roofline to reduce the frontal area
Smooth transition near the front fascia Corner rounding along
the sharp corner
Boat-tailing near the rear fender for a larger pressure recovery
Raising the cowl area to provide a smooth and continuous flow acceleration
The modified shape reduced the drag by 4%
Red: push outward
Blue: push inward
ADJOINT EXAMPLE RESULTS
ADJOINT SUMMARY
• Currently using to give guidance to engineers and designers
• Seeing reduction in windtunnel and CFD iterations
• Future integration with other MDO approaches
LIGHTWEIGHT MATERIALS
Philosophy nm mm mm m
log(Length scale)
Engineering
Materials
Science
Chemistry
Physics
Competency
Mg flowMg flow
Electronic
Atomistic
Microstructural
Continuum
$
MULTI-SCALE MODELING OF MATERIALS
Med. Mn (10wt%)
Q&P
ICME: 3RD GENERATION STEELSIntegrated Computational Materials Engineering
MathWorks tools used to measure
retained austenite volume fraction with
strain and parse data
PERSONAL SECURITY
CRASH TEST DUMMIES
INDIRECT INJURY
MEASURES:
• Head injury criterion
• Neck forces
• Chest deflection
• Femur load
• Tibia index
Crash
Simulations
Dummy Response
Crash Model
CRASH DUMMY BASED SIMULATION
HUMAN-LIKE CRASH TEST/SIMULATION TOOLS
Reduced Biofidelity
GLOBAL HUMAN BODY MODELS CONSORTIUMGHBMC
• Co-founded by GM in 2006, GHBMC is an international consortium of automakers & suppliers working with research institutes and government agencies to advance human body modeling technologies for crash simulations.
• OBJECTIVE: To consolidate world-wide HBM R&D effort into a single global efforts
• MISSION: To develop and maintain high fidelity FE human body models for crash simulations
MEMBERS
SPONSOR PARTICIPANTS
GHBMC DEVELOPMENT STATUS
• Developed 10 models
• Three more detailed pedestrian models (M50-P,
F05-P, and M95-P) to be delivered in 2017
F05-OSM50-O M50-OSM95-O M95-OSF05-O F05-PS M50-PS M95-PS6YO-PS
Detailed Simplified
BELTED
PASSENGER
UNBELTED
PASSENGER
MathWorks tools used for crash
sensing algorithm development
TISSUE LEVEL INJURY ASSESSMENT:
• Skull fracture
• Brain Injury
• Vertebra fracture
• Spinal cord injury
• Rib fracture
• Major artery injury
• Lung injury
• Liver injury
• Pelvis fracture
• Femur fracture
• ………
Human Body Response
Crash
Simulations
Crash Model
FUTURE: A HUMAN BODY MODEL BASED VEHICLE DEVELOPMENT PROCESS
24
FUTURE: INFORM FIRST RESPONDERS AND ER
VEHICLE PROGNOSIS
MOTIVATIONEverything wears out over time
Customer’s life is disrupted, when his/her vehicle needs repair unexpectedly
OnStarTM Proactive Alert – A new customer care service
– Alert before failure happens
– Transform an emergency repair to planned maintenance
– Enhance ownership experience - a delight to customers
PROGNOSTIC ALGORITHM DEVELOPMENT
34
Physical-model based algorithm generation:
• Study failure modes - FMEA
• Model physics of failure
• Generate fault signatures and failure
precursors
• Develop prognostics algorithm
• Validate concept on benches and test
vehicles
Lead Acid Battery (Plate Surface Scanning Electron Microscopy)
Electric Motor
MathWorks tools used for algorithm
development and data analysis
PROGNOSTIC ALGORITHM DEVELOPMENT
35
Big-data based algorithm validation:
• Collect data from >1M vehicles
• Analyze warranty return parts
• Correlate algorithm outputs with
engineering assessment
• Calibrate algorithm parameters
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
0.04
0.08
0.12
0.16
0.2
0.24
Crank Time To 200 RPM (s)
Perc
enta
ge
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10
0.2
0.4
0.6
0.8
1
1.2
Mem
bers
hip
Gra
de
Fuzzy Set Long
3
6
12/17 01/06 01/26 02/15 03/07 03/27 04/16 05/06 05/26 06/150.2
0.25
0.3
0.35
0.4
0.45
0.5
Date
Cra
nkT
imeF
orC
onventionalV
ehic
les (
Second)
12/17 01/06 01/26 02/15 03/07 03/27 04/16 05/06 05/26 12/17
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
Date
Cra
nk
Ratio (
)
Statistical analysis
Time series analysis
VALIDATION
36
06/15 08/04 09/23 11/12 01/01 02/20 04/110
20
40
60
80
Date
Rho
c ()
GY
R
G
G
Y
R
G
Replacement Date
LIDAR DATA
ARTIFICIAL INTELLIGENCE AND LEARNING
SUMMARY
• New mathematical models and methodologies are driving automotive
innovation in a variety of areas.
• These approaches have reduced the time to bring innovative solutions to
the market.
• Themes
• Model Integration
• Collaboration
• The future of mobility will rely on continued breakthroughs in
• Model integration
• Big data
Recommended